AI Supply Shortages Undercut the Bubble Case
What's New
The AI build-out lacks the speculative excess capacity that defined past bubbles, because demand is running ahead of supply. Jon Gray, President and Chief Operating Officer at Blackstone, argues this in a conversation with Antonio Gracias at TomorrowX Summit. Gray says spending on models at Blackstone's portfolio companies is up 20-fold. Supply lags because of entitlements, moratoriums, chips, power and financing. He contrasts this with the 1990s fiber build, when companies laid capacity first and hoped demand would follow. Investors should judge AI exposure by contracted demand and access to power.
Why It Matters
Gray's view challenges investors who treat AI capex as a repeat of the telecom overbuild. It also shifts the pressure to lenders. Lenders limit their exposure to any single theme or company. Gray therefore expects more of the build to be funded by hyperscaler cash flows and new equity. He also expects debt costs to rise. Data centers are Blackstone's biggest area of concentration, so a benign view of the cycle supports its own book.
By The Numbers
130 times: Cisco's forward earnings multiple at the dot-com peak. Gray uses it to show what excessive valuation looks like.
$1 trillion: the EBITDA, before capex and taxes, that the largest builders will generate next year, in Gray's estimate.
15 to 20 years: the typical lease Blackstone signs with data center tenants. Those tenants have an average market cap of $2.5 trillion.
70,000: Blackstone's estimate of workers on its US data center sites by the end of next year. A couple of years ago, the figure was about 10,000.
Memorable Quotes
"Betting on the electrification of everything to us seems like a really smart way to play the picks and shovels of this technological revolution." Gray puts power and grid assets at the start of the AI value chain.
"You're playing this mega trend, but you're doing it in a long-term safe contracted way." Blackstone gets its AI exposure through long leases to the largest tenants, which limits the downside if demand cools.
"It doesn't mean every player is well capitalized. And yes, we'll probably see a shakeout here." Gray concedes that weaker builders may fail. Elsewhere in the conversation, he says fewer competitors would help Blackstone.
"We don't have that dynamic where you're going to wake up one day and say, 'Wow, we have all this speculative excess capacity.'" This line states the thesis in a form that can be tested.
The Wrap
Blackstone's case depends on demand staying ahead of supply. It holds while model spending at operating companies keeps climbing and power, permits and chips stay scarce. It fails if supply catches up before productivity gains reach revenue, which would leave capacity without tenants. Financing is a second risk. If lenders' concentration limits raise debt costs faster than new equity fills the gap, smaller builders stall. Data center leasing and debt pricing through the end of next year will settle it.



Comments